pdf-tools | Emacs support library for PDF files | Code Editor library

Assuming you are using M1 Mac, you can solve this problem installing mupdf, swig, freetype and latest version of PyMuPDF:

You should not use -dNOSAFER, you should instead add files/directories to the permitted file reading list using the --permit-file-read switch. The file which needs to be read is the OutputIntent profile which is one of the main ingredients of the pdfa_def.ps file. See below.

If you do not include the pdfa_def.ps file then you will not get an OutputIntent in the final PDF/A file and it will not be PDF/A compliant (unless you specify UseDeviceIndependentColor as the ColorConversionStrategy). That's why your code example doesn't work. Noticing that PDF-Tools still says the file is valid, I would stop using that as a validator, it clearly isn't reliable. I've found VeraPDF to be the best validator personally (it's better than the Acrobat built-in verification).

I'm surprised that the command line you have shown at the top of the question produces a valid PDF/A file, unless you have modified the pdfa_def.ps file? You are supposed to and in particular you must modify the value associated with the /ICCProfile key. That value (a string inside parentheses) needs to be a fully qualified path to the ICC profile and either the ICC profile file or the directory it resides in needs to be added to the permitted list of files to read see the documentation here under -dSAFER.

Assuming you have done so, then the resulting PDF file should be a PDF/A-1b conformant file. And indeed according to your question, VeraPDF says it is conformant so I'm unclear on what your problem is there. It would be much more useful to share the input and output PDF files rather than a picture of (part of) the Acrobat display.

So to answer your questions:

1. Yes there is a way to generate a PDF/A file with ICC information (it isn't valid if it doesn't have an OutputIntent) and your command line does so. If you have not modified the pdfa_def.ps file appropriately you may still have a problem.

2. As far as I know you run the pdfa_def.ps file using Ghostscript.NET in exactly the same way as you do on the command line, you just put it in the list of arguments. So you just need to uncomment the line you've commented. Of course, you haven't included -dNOSAFER, nor added the ICC profile to the list of permitted files to read, so you will get an error. I am surprised you are getting a fatal error though, I'd expect an invalidaccess, but the obvious thing to do is to add -dNOSAFER to the arguments. The back channel output might be useful, it may have more information, and you haven't included that.

Oh, and I would not write to a pipe either. The pdfwrite device expects to be writing to a file and it may try to seek within the file while writing it. If it does and you've written to a pipe (or other non-seekable output), then it's going to fail.

You don't need to add -f to the argument list, and this:

It's a HTML encoding issue. I think you need to either use the following code.